MiniCPM-o 4.5: Towards Real-Time Full-Duplex Omni-Modal Interaction

Benchmark Model Rank Results
audio-visual-question-answering-on-futureomniMiniCPM-o 4.5-Instruct#1Accuracy: 56.1
video-question-answering-on-videoholmesMiniCPM-o 4.5-Instruct#2Accuracy: 64.3
visual-question-answering-vqa-on-mmt-benchMiniCPM-o 4.5-Instruct#2Accuracy: 69.7
audio-visual-question-answering-on-worldsenseMiniCPM-o 4.5-Instruct#3Accuracy: 55.7
audio-visual-question-answering-on-dailyomniMiniCPM-o 4.5-Instruct#4Accuracy: 80.2
text-to-speech-synthesis-on-seed-tts-test-zhMiniCPM-o 4.5#4CER: 0.86
visual-question-answering-vqa-on-hallusionbench-average-accuracyMiniCPM-o 4.5-Instruct#11Accuracy: 63.2
multi-task-language-understanding-on-mmluMiniCPM-o 4.5-Instruct#14Average (%): 77.0
text-to-speech-synthesis-on-seed-tts-test-enMiniCPM-o 4.5#17WER: 2.38
optical-character-recognition-ocr-on-ocrbenchMiniCPM-o 4.5-Instruct#18Score: 876
video-question-answering-on-mlvuMiniCPM-o 4.5-Instruct#18Accuracy: 76.5
zero-shot-video-question-answer-on-video-mme-w-o-subsMiniCPM-o 4.5-Instruct#25Accuracy (%): 70.4
arithmetic-reasoning-on-gsm8kMiniCPM-o 4.5-Instruct#28Accuracy: 94.5
vision-language-understanding-on-mmstarMiniCPM-o 4.5-Instruct#29Accuracy: 73.1
multimodal-reasoning-on-mathvistaMiniCPM-o 4.5-Instruct#30Accuracy: 80.1
video-question-answering-on-lvbenchMiniCPM-o 4.5-Instruct#34Accuracy: 50.9
code-generation-on-mbppMiniCPM-o 4.5-Instruct#41Accuracy: 76.7
code-generation-on-humanevalMiniCPM-o 4.5-Instruct#42Pass@1: 86.6
multimodal-reasoning-on-mmmuMiniCPM-o 4.5-Instruct#44Accuracy: 67.6
mathematical-reasoning-on-math-500MiniCPM-o 4.5-Instruct#102Accuracy: 77.0